remove enrollmentComplete experiments from pools unless already enrolled

[bcb37]

No description provided.

[Copilot]

Pull Request Overview

This PR removes experiments in ENROLLMENT_COMPLETE state from experiment pools unless the user is already enrolled in them. This prevents new users from being assigned to experiments that have completed enrollment while still maintaining existing enrollments.

• Filters out enrollment complete experiments for non-enrolled users while preserving them for already-enrolled users
• Updates processExperimentPools method to accept exclusion parameters and handle enrollment state filtering
• Adds comprehensive test coverage for the new filtering behavior across various scenarios

Reviewed Changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
ExperimentAssignmentService.ts	Main logic change to filter enrollment complete experiments and refactor pool processing
ExperimentAssignmentService.test.ts	Extensive test coverage for new filtering behavior with multiple test scenarios
WithinSubjectEnrollment.ts	Remove unused import and function call
Scenario2.ts	Simplify test expectations for user 4 assignment behavior
RevertToCondition.ts	Simplify test expectations for user 4 assignment behavior

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[bcb37]

Supporting multiple assignments for a single decision point in the pooling logic will result in unpredictable behavior in at least the js client library, since that's one of possible multiple places that rely on the assumption that there will be a unique assignment for each decision point in non-within-subject experiments.

[danoswaltCL]

I have been looking at this btw, I went down a rabbit-hole yesterday using your branch to investigate how we employ this logic in our no-experiment-id scenarios in mark. It seems like either do caching or try to shoehorn this logic into mark, which may or may not have all the info necessary to do it at the moment, so would need some rejiggering. I looked into caching and it honestly seems more straightforward than other refactor options. Both options seem difficult to employ by Friday...

[danoswaltCL]

ok, I have stop looking at the experiment assignment service it's got me spinning. any further questions I have are really outside the scope of this change, this looks good.